New Prescription Formula Brings Precise Fitting to Skin-Worn Bone Conduction Hearing Devices

For babies born with ear canals that never formed, or children whose middle ears repeatedly fail them, sound reaches the brain by a different route: vibration delivered directly through the skull to the cochlea. Bone conduction devices have made this possible for decades, but fitting them accurately has lagged far behind the precision achieved with conventional hearing aids. A team of Canadian audiologists and software engineers at Western University’s National Centre for Audiology, working with colleagues at the University of Alberta and Drexel University, has now released Version 2 of the Desired Sensation Level Bone Conduction Device prescription, or DSL BCD v2, a computational formula that for the first time brings evidence-based, objectively verifiable targets to the skin-driven devices that infants and young children overwhelmingly depend on.

The significance of the update lies in who it serves. The earlier version, DSL BCD v1.1, was a genuine breakthrough when introduced in 2017 by William Hodgetts and Susan Scollie, but it only worked for direct-drive devices attached to a surgically implanted abutment that penetrates the skin. Those devices suit adults and older children who meet skull thickness requirements. Very young children, however, cannot undergo such surgery, so they wear non-invasive transcutaneous devices held against intact skin by soft headbands, hard headbands such as the Cochlear SoundArc, or adhesive adapters like the Med-El ADHEAR. Because v1.1 assumed the patient could complete in-situ audiometry through their own abutment, clinicians had no validated prescriptive pathway for the tiniest patients, precisely the group for whom consistent access to amplified speech matters most for spoken language development.

The physical problem the new version tackles is the skin itself. When a device vibrates through skin and soft tissue rather than through bone directly, energy is lost along the way, and the loss is not uniform across frequencies. Classic studies comparing percutaneous and transcutaneous transmission found attenuation reaching up to 28 dB at higher frequencies. More recent work by Gascon and colleagues in 2022 and by El-Naji and colleagues in 2024 and 2026 compared thresholds measured with a standard audiometric bone oscillator against thresholds measured through the user’s own skin-worn device. Within each listener, the in-situ thresholds were consistently worse, by amounts that were frequency-dependent and device-specific, with shifts of up to 40 dB noted at the highest frequencies. Contact force, contact area, and the transducer’s reference equivalent threshold force level all contribute to this variability.

DSL BCD v2 absorbs this complexity through a new audiometric-to-dial-level, or ADL, correction. The idea is elegant: audiometric bone conduction thresholds, obtained with standardized clinical transducers, are shifted by lookup-table values to predict the thresholds the patient would show if tested through their own skin-driven device in place. The corrections differ for adults and children, and for pressure-worn versus adhesive devices, because the underlying evidence comes from distinct populations and because young children’s test results carry developmental quirks. Pediatric data came from a retrospective review of clinical files from three clinics following a common protocol, involving 29 ears fitted with six different brands of skin-worn devices, with roughly half tested using visual reinforcement or conditioned play audiometry. Adult corrections came from studies of 20 young adults wearing pressure-fitted devices and 36 young adults wearing adhesive devices.

The pediatric correction carries particular weight for early intervention. Children tested with conditioned procedures often respond at a minimum response level rather than a true threshold, larger step sizes bracket their responses, fewer frequencies can be measured within a session, and electrophysiological estimates in nHL require conversion to estimated hearing level. The v2 workflow accommodates standard audiometry, pediatric audiometry, and auditory brainstem response inputs, each with its own ADL values, rounded to the nearest 5 dB to match audiometric resolution. Differences about the mean ran roughly 10 dB across frequencies, with a notably larger correction at 4,000 Hz, reflecting the known weakness of skin-driven output in the high frequencies. The corrections were calibrated against audiometers using the pre-2025 ANSI reference equivalent threshold force levels, and the team has derived adapted values for clinics already transitioned to the new ANSI S3.6-2025 standard during what they describe as a temporary bridging period.

Once thresholds are established, the algorithm preserves the conceptual machinery that made the original DSL method successful for air conduction hearing aids. Thresholds in dial level are converted to decibels of force level through a published dial-to-force-level transform, and a real-head-to-coupler-difference, or RHCD, can further map values between a skull simulator and the on-head reference point. Because device-specific RHCDs did not exist for skin-worn devices, the team compiled manufacturer-provided reference values for 17 devices from Cochlear and Oticon Medical, computed brand-specific medians, and adopted the midpoint as a generic transform. The upper limit of the patient’s auditory area is set by whichever is lower, the measured or predicted loudness discomfort level or the device’s manufacturer-specified maximum force output, then compressed according to bone-conduction loudness growth functions.

All of this converges on the FLogram, a force-level analogue of the SPLogram display familiar from hearing aid fitting, now available in commercial verification equipment. Prescribed targets for soft, conversational, and loud speech are plotted alongside the patient’s thresholds, comfort limits, and measured device output, allowing clinicians to fine-tune a bone conduction device in a hearing aid test chamber much as they would a conventional hearing aid. The prescription distinguishes adult and child target types, adjusts for listening in noise with a higher compression threshold and modified frequency importance weighting, and reduces bilateral adult speech targets by 3 dB. Verification targets can be generated for either a skull simulator or on-head force measurement, and the software tags the coupling type, abutment, pressure, or adhesive, so that patient records remain interpretable across transitions from skin-drive to surgical fittings.

Simulated case examples, run on the Audioscan Verifit 2 system with a Baha 6 Max and a Ponto 5 Superpower, illustrate the workflow in practice. One simulated 12-month-old with bilateral atresia, fitted at two months of age from ABR thresholds, was reprogrammed at twelve months using in-situ thresholds obtained through visual reinforcement audiometry. A simulated 24-month-old with mixed hearing loss could not comply with in-situ testing, so conventional VRA thresholds were entered instead, with the pediatric ADL correction generating targets matched to within 5 dB root-mean-square error between 500 and 3,000 Hz, though device output limits prevented meeting targets above 3,000 Hz. The team emphasizes that choosing devices with higher maximum output matters: comparing an older and a newer processor on identical thresholds produced nearly identical soft-speech targets but a 7 dB difference for loud speech, because the smaller dynamic range forces more aggressive compression.

The authors are candid about uncertainty. Every correction represents an average drawn from modest samples, and individual skin thickness, headband tension, skull maturation, and audiometric step size scatter real values around those means. Errors from threshold estimation, ADL prediction, coupling effects, and RHCD transforms can accumulate, with the error from assuming an average ADL ranging from 10 to 20 dB depending on frequency and population. Their deliberate safeguard is conservative: ADL corrections are applied fully at threshold but not to loudness discomfort level predictions, reducing the risk of over-amplifying loud sounds in children. Wherever possible, they recommend measuring in-situ thresholds directly, reserving ADL corrections for situations such as infant electrophysiology where in-situ testing is impossible.

Validation remains the essential next step. The prescription now needs field testing to confirm that its targets deliver comfort, speech intelligibility, and acceptable loudness in real-world use, as has been done extensively for air conduction formulas. Future work may add age-specific or individually measured RHCD values, a microphone location effect for forehead placement sometimes needed with tiny infants, and eventually a strategy for transcutaneous induction implants. For now, the release of DSL BCD v2, implemented as a dynamically linked library embeddable in commercial fitting and verification systems, means that the smallest bone conduction users, the infants and toddlers in their most language-critical years, can finally be fitted with the same objective, measurable rigor long standard for everyone else.

Subject of Research: A prescriptive formula and normative correction data for fitting and verifying skin-driven bone conduction hearing devices in children and adults

Article Title: The Desired Sensation Level bone conduction device prescriptive formula: new adaptations and applications for skin-driven bone conduction fittings

Article References: Scollie, S., Beaulac, S., Beh, K., Holden, M., Bagatto, M., Banerjee, S., & Hodgetts, W. B. (2026). The Desired Sensation Level bone conduction device prescriptive formula: new adaptations and applications for skin-driven bone conduction fittings. Frontiers in Audiology and Otology, 4, Article 1846344. https://doi.org/10.3389/fauot.2026.1846344

Image Credits: AI Generated

DOI: 10.3389/fauot.2026.1846344

Keywords: bone conduction devices, pediatric audiology, DSL prescription, hearing aid verification, skin-drive coupling, audiometric thresholds, force level, skull simulator, conductive hearing loss, early intervention, audiometric calibration, prescriptive targets